xemdetia

I think this feels like a misunderstanding of what the PPP loans were designed to do. They were an economic incentive to keep people on payroll instead of a mass layoff or long term furlough. The correct business decision if you are no longer allowed to operate a business or you are no longer bringing in revenue to support the business you currently have is to do layoffs. If you have anticipation that the revenues will return to normal you can try a furlough. There's no economic incentive to keep people employed if all expenses including payroll suddenly eclipse the revenues of a business, and there's no incentive for a business owner to put personal wealth into their business if it has a great chance of never recovering.

The fact that the fixed expenses during this pandemic did not stop for corporations made the expense of payroll which provided significantly reduced revenue stand out. Even law offices and physicians have very high recurring expenses like insurance that they just can't stop and start on a dime and stay in business. It made no sense to keep people on the books until PPP gave a compelling reason to do so. It isn't about 'keeping a wealthy lifestyle' as much as 'keeping a functioning business' a lawyer or physician with no practice is just as much a drag on the economy as anyone else.

The infinite scroll problem I feel comes up most that if there was a useful piece of information in the middle of an infinite scroll it is likely impossible for you to return there with any sort of reliability. You can't save the URL because it probably doesn't encode anything useful about where you are in a most cases. You can't logically organize 'where' this information is because its location is pure happenstance. It is pointlessly ephemeral. I can't start on page 5, I can't start on page 100. I have to scroll and just pray I get to get lucky which is the function of a slot machine, not someone who's trying to present me information. If I _really_ wanted to get that much more content I am more than happy to click the next button.

Yes, it really is getting out of hand. If there was a reasonable equivalent that was an alternate (like an email address used to be), that would be one thing but even with a US phone all of a sudden your phone dies and you are completely stranded and even sites you used to be able to access are now no longer accessible because they are asking for phone numbers and confirmation of said numbers. It feels really awful.

I feel like I didn't get the same satisfaction out of factorio until I turned the difficulty up enough to feel like the bugs were a threat so I could lose. Admittedly I turned it up too high to get that feeling and had to restart with a bit less difficulty. Then I had a reason to build walls/turrets/etc because I couldn't maintain safety without them and planning for mining expansions.

I'm really hoping that gopls gets to 1.0 soon. I feel like gopls is the elephant in the room that's taking a lot of air from some of the smaller tools, and I'm glad it's being called out as a priority in this report.

One of the primary methods to join multiple learned models is a decision tree (or combination of decision trees making a decision forest), which can be simplified as a series of if statements/conditionals. So if you join a 'is it a shirt' model with a 'is it striped' model you get two sets of things, and with how big data approaches this it is something you can do quickly. As other people have pointed out the issue here is that the NLP of the actual search is not creating a negation of two sets, it is returning the intersect of sets 'is it a shirt', 'is it striped' and shrugging its shoulders and either intersecting it with 'things with the text without', or throwing up its hands entirely based on the context because it wasn't programmed to do something smarter.

The drones have always felt like a quixotic adventure for GoPro that seemed like a weird distraction for other product improvements.

I don't even think that GoPro should be classified as part of the adventure usage category anymore, as most of the GoPro usage I have seen has more to do with the fact that it's a compact camera that's relatively durable or sacrificial depending on your application. What I've been seeing over the last few years is that GoPro cameras are less handy than a smartphones whose recording capabilities just keep enhancing from the last few generations of improvements seemingly primarily focused on that, especially as more people have been streaming video directly to viewers from their phone compared to a recording, edit, and then upload workflow. A few years ago I felt like I saw GoPro's being used for most of the content in a particular edited video but nowadays I see them used for one or two shots where the camera was more at risk. If anything I just see them getting pushed from mass appeal to a niche compact camera market because there's just better competition in this space and the middle to high end of the GoPro line competes with some of the entry level/second hand Canon and Nikon cameras for recording video and unless you have those space and durability requirements in the niche where GoPro shines.

I started this way and my next step was taking advantage of the current package manager. The fact you can inject new functionality so quickly made me really fall in love with emacs. Adding additional programming modes/file format modes without any headaches has been a boon to my usage of Emacs. I'm rarely tuning settings in elisp anymore and instead just adding a feature so the file has colourization and tab stops working properly. If I do actually want to customize something since customize-mode works well enough now I can usually find the modest setting I do want to change without much hassle.

I use a similar methodology when I write as I am often writing technical documentation and not as much fiction these days. In technical writing it is easier as there's a fundamental understanding that any base concept that's involved must be defined and explained before it is applied. This introduces a natural dependency chain of things that have to happen and can be used to help sketch out a basic order. Since you have the constraint of linear top to bottom text this often at least makes ordering beginning, middle, and end talking points obvious.

If you are writing a technical document ordering of points like this tends to look like an outline format but the dependencies are not visible in the end product and are often implied.

If you are writing about a report of events/fiction there's often also natural checkpoints or serialization paths. If Alice and Bob go to a casino at the same time, there may be a logical ordering that makes sense to finish some of Alice's and Bob's story individually with them both arriving at the casino. This allows the reader to unload the idea of the action of 'arrive at the casino' and use that as the foundation for new ideas without having to reconsider it. These new 'checkpoints' are places you can refer to, as in 'this thing happened because of how Alice and Bob came to the casino' without re-explaining everything. In Math you can refer to a previous Lemma, in technical writing you can refer to a previous section or figure. This concept also helps the idea that people can only generally keep a handful of things at the top of their head at once and can help collapse multiple individual ideas into a more common concept (similar to refactoring 'extract function' in programming).

So in actual practice this benefits by using a media or tool that's easy to rearrange sections and make dependencies. For physical media I've used blank index cards and for tools I use Emacs and org-mode because I can quickly rearrange outline headers. In this outline I list dependencies to other outline headings and external documents. This should only talk about the structure and not really go into things you'd like to write there, and generally by working through the dependencies you can see how you get from idea a to b to c and can lead the reader exactly where you want them to go. I can see how using a mind mapping software or draw.io to build a flowchart a good method as well but the act of moving things and reordering things needs to be cheap and intuitive to your thought process.

Another way I often approach more creative assignments is figuring out where you want to end up and working backwards until you get where you want to start. Much like going from the end to the start of a maze is often easier I approach writing the same way. It also helps make sure you have a direct line from the first idea to the end idea at least on paper as you assemble what's important. The end here can vary depending on what kind of document I'm writing whether it's a final statement of an argument, a single idea or multiple set of ideas I want to explain to someone, or the climax to a story. The way you find the introduction and start point this way is by thinking of your target reader and what you need to give them to get them naturally and easily to the ending you've already decided on. Because you know where you're going to end up it's easier to manage fanning out to all the supporting ideas that need to get to that end idea without bringing in things that are irrelevant or actively distracting. Because you are thinking about it like a flowchart instead of an outline, it's OK at that point to fan out dozens of ideas, and then fan out the precursors of those supporting ideas, and then fan out the precursors of those ideas until your target reader if they read everything would be able to comfortably reach your final idea. Then you can start doing some of the dependency and checkpoint work that I described earlier and the ordering starts to emerge from the pile. I find this method also helps my writing be more targeted towards the goal and does a better job engaging the reader (where this post is a poor example).

Thanks for putting together this in a book, I feel like I've written this book multiple times for compliance and explained the concepts but sadly that doesn't cross employment barriers. Also annoyed that I didn't make the time to write a lesser variant of this book first. I just hope this book's PDF name doesn't make my new title 'SRS' on the resume going forward.

Yeah, this is the kind of way I considered this article. There's so many operations including medical hospitals purchasers that are trying to follow the current mantra of as little idle inventory and wastage of inventory on the shelf as possible for as cheap as possible. This makes sense for people purchasing finished goods but doesn't make as much sense for people making finished goods who may need primary material goods in quantities and timings that do not scale well in increments. It also makes no sense from a machining and tooling efficiency investment to replace a particular operation with a machine to do it 100x as fast if making 100x of the particular sub-assembly completely overwhelms the maximum capacity of the next step unless it is also improved.

There's just a minimum volume cost to most finished goods below which it just costs money to do the work at all. Sure if the volumes go way up there's economy of scale at work and the fixed costs of scaling up go down so prices can be reduced in a way that doesn't affect the manufacturing business to be more competitive but there's a bottom in the other direction.

You can insulate that with long term contracts from the buyers so that you can get favorable contracts with the suppliers, but when you have purchasers going quarter to quarter and saying lowest price = best value you just are playing a totally different game for a smaller manufacturer. The current model of a small American manufacturing business seems to currently reward doing the same boring product at the same boring volume so that the company will exist in 20 or 30 years by investing in specialist equipment that does one or more steps of manufacturing better than generalist equipment and creating a niche stable market for yourself as the depreciation/fixed cost of that specialist equipment per total units made approaches next to nothing.

I may just be old and bitter but I've never found emojis to actually help in commentary and are actively distracting when people already have established norms in communicating in a particular language as to my eyes it becomes adding a second language where there was already one. There's no rock solid definition for any emoji that has been normalized with a universally accepted dictionary especially when it is intermixed in another language that has its own complex grammar rules. As a single character statement of a symbol e.g. 'thumbs up' it works well, but that's the limit of emoji usefulness.

As someone who works in tech I'm always aware of the hundreds of completely overloaded terms I have to use and redefine on the fly but I also have the structure of the language I am using to provide context clues to have my actual intent, meaning, and purpose of said writing last to any reader including myself five years from now. I have so little confidence in effectively using emoji to actually provide expression in a way that is persistent to myself in five years that I abstain from it entirely. In small communities of people I have seen emoji-esque symbols exist (slack/forums/odd IRC emoticons) but it is because that is part of the living culture of a particular community and has to be taught to each person that chooses to join a community. I feel that is fine because there is a context to it and an established rule of use that allows it to enrich the conversation, but on a site like HN where it's a lot of people coming together it seems like trying to put a wall up for communication. A simple example is well is most HN commenters fully understanding what a line that starts with '>' usually referring to a quote of a prior post that they want to highlight in particular. It is a useful shorthand compared to writing out longhand 'In your post you have written many things but I particularly want to respond to your comment that reads "x to y".' You have to learn that and it's overhead.

I also depending on theming/font of the emoji characters themselves can sometimes take the wrong tonal interpretation even to me. Some seem sarcastic, some seem more intimate, some seem hostile, even though the underlying emoji is supposed to be neutral in tone. This is another reason why I avoid them, as I don't want to give off the wrong impression based on that.

For a discussion forum like HN I'd say that emojis are generally harmful to making sure that the conversations last longer than the moment they were written in case it becomes useful tomorrow.

For Intel chips i3/i5/i7 have generally represented feature ranges, so to really identify the chip you have to find out their product code or at least processor generation. Once you have that you can use ARK to distinguish between them. Here's an example of an older 2017 era desktop processor of i5 and i7: https://ark.intel.com/content/www/us/en/ark/compare.html?pro...

Complicating things recently is the introduction of the Xeon W and i9 lines, but for the most part the Xeon ones have yet more a new features and i9 are higher core counts to compete with AMD.

In broad strokes in the laptop space the majority of enhancements since 2013 I would say is higher RAM capacity (that Apple as a manufacturer has not to pursue compared to Lenovo, etc.), still yet lower overall power consumption to improve battery life, DDR3 to DDR4 ram, bigger L1/L2/L3 caches, and normal improvements to embedded graphics to support bigger onboard and external displays, and PCI-E enhancements for drives as NVMe-style drives are what I would consider the new standard.

I would say that it depends on the role when it comes to system design and may be reflective of your level of experience. Certain roles require you to know how to run something in a production context because otherwise the dev will make bonehead decisions. It's important to interview for that aspect as well for growth as there are some people that enjoy working on problems but just are awful at maintainable system design or conversely consuming large complex base systems. Anyone who 'develops for cloud' generally needs more system design skills just to understand how to work and build in cloud compared to a developer where everything is provided for them when it comes to actually delivering the code. This is a large change as well for people playing containers, kubernetes, and so forth where a lack of system design fundamentals can be a direct impediment to actually doing something of use given the constraints.

I think it was funny to presume that the employer wouldn't pay for it. At one of my old roles I had a second laptop I used solely for VMs just because I ran out of space on primary workstation. A simple fix orders of magnitude cheaper for a developer compared to more ESXi or cloud hardware.

I wish this data was sliced down better into full time and part time work and/or hourly and exempt jobs. I feel there are more 'bad part time' jobs now than in years past that don't allow someone to give appropriate notice without reprisal.

I agree with the concept that it is turning into something else, as I see DevOps/SRE/CI/PM people doing QA work through trying to just make the thing go. It just is leaving a lot of gaps in longevity, complex setup, and building tests either manual or automated to do that better. In the last dozen or so teams I've interfaced with this is just getting worse as I'm seeing lots of situations where things like backup/restore breaks from the last hotfix, upgrade a system a few sequential versions, and only covering X-1/X critical features in ordinary testing. These are things that a traditional QA team would cover after it burns the business/org in one way or another, but since it's not a core focus of some of these people covering parts of the same role it's leaving a lot of not-obvious gaps in a lot of groups.

As far as I have seen there is still updating overhead where you have to initiate upgrades to a known stable/supported version at a regular cadence. EKS suggests having an E2E CI pipeline to test it before updating production, and I feel like that's the only way to do it. There is nonzero churn even though it's managed.

Yeah as someone familiar with openssl it looks like a version of openssl that was built incorrectly.

I always find myself thinking of articles like this when these collisions come up: https://www.theatlantic.com/magazine/archive/2019/07/future-...

It always seems to me that modern staffing modeling/logic just skips emergency situations especially for critical systems. When I read this idea that generalization and reduced staffing was to replace specialization for exactly this reason. It's hard to moonlight as an electrician during the worst moments, even if you can somehow skimp by at the best moments. The whole point of a navy vessel is to be as self sufficient in all conditions not to have to page a vendor halfway around the world just because the system blew up again, and that's just orthogonal to a business where you can put in a few extra hours to fix a problem or can wait until Monday for a vendor to call back. I'd be interested to know if some people have seen better operations research and modeling to cover these sorts of organizational systems compared to the apparent style of cut a labour force until you see blood and reap the profits from minimizing that expense against long term stability.

Changing the search engine that powers the omnibox is different than changing the behaviour of the omnibox, which is kind of a lot when it comes to predictive search, predictive completion and a bucket of other things that I wish you could turn off (or can be turned off for a while before Google decides you shouldn't ever need or want to turn it off). It's a target with a huge amount of churn. I agree that Firefox is a more sane target as a developer.

KVM and libvirt also support GPU pass through.

I agree that the hotel prices definitely drive people out, but also the general logistics problems, and things to do that aren't a conference are really silly in San Francisco compared to holding it in other more tourist-focused areas like Las Vegas. Las Vegas is capable of holding simultaneous high headcount events, and has the infrastructure to get a large number of tourists around without even batting an eye.

I'd be interested to hear other people's experiences but it feels like most conferences in San Francisco would start to become miserable in hotel price and commuting for people coming from out of town once you start getting to 20,000 or more people.

I actually use emacs git-gutter to have a more visual way of seeing what's going into a commit or not. Having the inline annotations in the files I am working on it is great for doing the staging parts of a hunk.

Have you looked into VLC or OBS for doing this? The big issue that isn't really solved is that most desktop protocols take advantage that the image has a majority of the screen static and only small areas of the screens need updates. Video is the exact opposite problem: nearly every pixel changes across frames and so to have it be reasonably well sync'd especially with sound you need a different kind of protocol.

https://www.videolan.org/vlc/streaming.html

You can also buy a business laptop with an embedded Quaddro and have the same sit in the chair experience at that moderate graphics level without much difficulty. In a lot of ways some of the recent generations for Intel embedded graphics have been baseline requirements for many games (mostly driven by F2P and MMO games) and you don't even need a real GPU just to play vs. having a AAA ultrahigh setting experience. I mean if I can beat Dark Souls on a $300 staples laptop and enjoy the ride I don't see where these grandiose requirements are coming from.

If there were more PC exclusive games that were truly GPU required monsters I could see the point of Stadia, but as time has progressed most AAA games that end up targeting PC usually have a generous minimum requirement threshold, and often its so generous that most things can actually do the job. The only place where the GPUs have been really required in the last few years is the VR stuff which because of general latency of input/response for Stadia would probably cause a lot more motion sickness in people than a local system. Stadia's benefit is capturing the kind of user that doesn't care about quality (because if you did you'd just buy that tower), but those are the kind of users that would make do with what they already had anyway for cheap.

In my opinion this feels like the same slow death as OnLive, with even less benefit in knowing that OnLive's entire focus was on streaming gaming rather than Google who would put this product on ice in 3 years.

My main issue for Stadia has been that they are trying to rebadge existing games to work on Stadia instead of making a game that uses the cloud to provide a unique experience. It's one of the things that none of the other streaming companies I am aware of have tried to make an advantage of the proximity of the actual VMs/shared runtime to make it a Mainframe-style experience where you wouldn't have to compromise as heavily of number of tracked objects in concurrent simulations as the simulations are being run spatially closer so the base latency between sims are wildly different than the current distributed model.

Movies are an easy example: Do you want to watch the original ending of Return of the Jedi, the special edition ending, the revised ending post-prequels, the 4:3 edition or the widescreen edition? There are plenty of reasons to want to keep a copy of the version of the experience you enjoyed and want to experience again.

I bought a HP small b&w office printer in 2005. I still am using the original toner I got with it then and it's still doing a fine job. I loved it so much I bought a few more to use in the office I was IT'ing for at the time, and saw them die glorious deaths after printing 10's of thousands of pages. Meanwhile I haven't seen or used a consumer inkjet printer that has lasted more than a few months.

Putting the cost on manufacturers and potentially breaking legitimate resellers just feels straight out of the Walmart strongarm playbook. It feels like it should be against some FTC regulation somewhere that a transferrer of a good has the responsibility to not allow this to happen. The only historical equivalent I can think of is the movements of crude oil through pipelines where if you sell one type of oil and get a different oil on the buyers end the pipeline owner doesn't get to wash their hands of the affair and say 'its just oil, buyer beware.'

Amazon should be subsidizing 100% of this program if they can't take care of their own mess.